Crenothrix are major methane consumers in stratified lakes
Kirsten Oswald1,2,6, Jon S Graf3,6, Sten Littmann3, Daniela Tienken3, Andreas Brand1,2, Bernhard Wehrli1,2, Mads Albertsen4, Holger Daims5, Michael Wagner5,
Marcel MM Kuypers3, Carsten J Schubert1 and Jana Milucka3
1Department of Surface Waters—Research and Management, Eawag, Swiss Federal Institute of Aquatic Science and Technology, Kastanienbaum, Switzerland;2Institute of Biogeochemistry and Pollutant Dynamics, ETH Zurich, Department of Environmental Systems Science, Swiss Federal Institute of Technology, Zurich, Switzerland;3Department of Biogeochemistry, Max Planck Institute for Marine Microbiology, Bremen, Germany;4Department of Chemistry and Bioscience, Center for Microbial Communities, Aalborg University, Aalborg, Denmark and5Division of Microbial Ecology, Department of Microbiology and Ecosystem Science, Research Network Chemistry meets Microbiology, University of Vienna, Vienna, Austria
Methane-oxidizing bacteria represent a major biological sink for methane and are thus Earth’s natural protection against this potent greenhouse gas. Here we show that in two stratified freshwater lakes a substantial part of upward-diffusing methane was oxidized by filamentous gamma-proteobacteria related to Crenothrix polyspora. These filamentous bacteria have been known as contaminants of drinking water supplies since 1870, but their role in the environmental methane removal has remained unclear. While oxidizing methane, these organisms were assigned an ‘unusual’ methane mono-oxygenase (MMO), which was only distantly related to ‘classical’ MMO of gamma-proteobacterial methanotrophs. We now correct this assignment and show thatCrenothrixencode a typical gamma-proteobacterial PmoA. Stable isotope labeling in combination swith single-cell imaging mass spectrometry revealed methane-dependent growth of the lacustrineCrenothrixwith oxygen as well as under oxygen-deficient conditions. Crenothrix genomes encoded pathways for the respiration of oxygen as well as for the reduction of nitrate to N2O. The observed abundance and planktonic growth ofCrenothrixsuggest that these methanotrophs can act as a relevant biological sink for methane in stratified lakes and should be considered in the context of environmental removal of methane.
The ISME Journal(2017)11,2124–2140; doi:10.1038/ismej.2017.77; published online 6 June 2017
Introduction
Freshwater lakes represent large natural sources of methane and contribute more to methane emissions than the oceans despite their comparably smaller area (Bastviken et al., 2004). Highest rates of methane removal are usually measured at the oxyclines, either in the water column or in the sediment. Lake Rotsee and Lake Zug in Central Switzerland are typical examples of temperate lake systems with methane fluxes across the oxycline of 13 ± 3 mmol and 10 ± 3 mmol m−2d−1, respectively (Oswald et al., 2015, 2016). Both lakes are stratified, with methane-rich hypolimnia, but whereas the shallow Lake Rotsee overturns annually, the deep Lake Zug remains stratified throughout the year.
In both lakes, the vast majority of the upward-diffusing methane is removed at the base of the oxycline atin situoxygen concentrations in the low micromolar range (Oswald et al., 2015, 2016).
Methane oxidation at the oxycline was shown to be coupled to the reduction of residual or in situ-produced oxygen, but there were also indications for methane-oxidizing activity under oxygen-deficient conditions (Oswaldet al., 2015, 2016).
Abundant gamma-proteobacterial methane-oxidiz-ing bacteria (gamma-MOB) were shown to be involved in methane removal in both lakes (Oswald et al., 2015, 2016). Gamma-MOB are considered aerobes requiring oxygen for methane activation, even though some cultured representatives can perform methane oxidation under denitrifying con-ditions (Kits et al., 2015a, b). Environmentally relevant representatives of gamma-MOB in lakes and other freshwater habitats belong to the‘classical’
genera of Methylobacter, Methylomonas, Methylo-sarcina and Methylomicrobium (Boschker et al., 1998; Bodelier et al., 2013; Oshkin et al., 2015), and all possess particulate methane monooxygenase
Correspondence: J Milucka, Department of Biogeochemistry, Max Planck Institute for Marine Microbiology, Celsiusstrasse 1, Bremen 28359, Germany.
E-mail: jmilucka@mpi-bremen.de
6These two authors contributed equally to this work.
Received 4 November 2016; revised 13 April 2017; accepted 21 April 2017; published online 6 June 2017
www.nature.com/ismej
(pMMO) as the key methane-oxidizing enzyme (Bowman, 2005). In Lake Rotsee and Lake Zug, unicellular gamma-MOB represented a stable com-munity at the oxycline. The bacteria showed rapid growth on methane as evidenced by the increase in cell abundances and the uptake of13C-methane into their biomass (Oswaldet al., 2015, 2016).
In these studies, gamma-MOB were identified by fluorescence in situ hybridization using the 16S rRNA-targeted oligonucleotide probes Mgamma84 +705. Interestingly however, these probes do not bind to members of a potentially important subgroup of gamma-proteobacterial MOB, the putative family Crenothrichaceae. Contrary to ‘classical’
MOB, these gamma-MOB are multicellular and filamentous. So far, only two of these bacteria have been documented in literature, Crenothrix polyspora and Clonothrix fusca, and both were retrieved from groundwater (Stoecker et al., 2006;
Vigliotta et al., 2007). Sporadically, environ-mental occurrence of Crenothrix is reported in literature based on retrieved 16S rRNA or pmoA sequences (Dörr et al., 2010; Drewniak et al., 2012), but its role in methane cycling has remained unclear.
The metabolism ofCrenothrixhas been a matter of debate since its first description as ‘Brunnenfaden’
(‘a well thread’; Cohn, 1870). Initially, Crenothrix/
Clonothrix filaments were considered to belong to the ‘iron bacteria’ due to the presence of metal particles in their sheaths (Roze, 1896; Jackson, 1902;
Molisch, 1910). This belief was challenged by studies that failed to observe iron encrustation in Crenothrix/Clonothrixfilaments (Kolk, 1938; Wolfe, 1960), and the later discovery of membrane invagi-nations has prompted suggestions for a methano-trophic lifestyle (Völkeret al., 1977). Eventually, the capacity to oxidize methane was experimentally confirmed on filaments retrieved from man-made habitats (Stoeckeret al., 2006; Vigliottaet al., 2007). Interestingly, C. polyspora was reported to possess an ‘unusual’ pMMO, which was only distantly related to ‘classical’ MMO of gamma-proteobacterial methanotrophs (Stoecker et al., 2006), and has now been recognized to cluster together with the ammonium monooxygenases of completely nitrifying ‘comammox’ bacteria (Daims et al., 2015; van Kesselet al., 2015).
Here we investigated the occurrence and involve-ment of these filainvolve-mentous bacteria in methane oxidation at and below the oxyclines of Lake Rotsee and Lake Zug. We performed stable isotope labeling experiments followed by single-cell imaging to explore the role of these microorganisms in environ-mental methane cycling, and metagenomic analyses to investigate their metabolic potential with respect to aerobic and anaerobic respiration. For compar-ison, we also performed metagenomic analysis of a sample from Wolfenbüttel waterworks sand filter reportedly containing high proportions of C. polyspora.
Materials and methods
Geochemical profiling in Lake Rotsee
Profiling was done in October 2014 at the deepest point (16 m depth, 47°04.259‘N, 8°18.989‘E).
A multi-parameter probe was used to measure photosynthetically active radiation (PAR; LI-193 Spherical Underwater Quantum Sensor, LI-COR, Lincoln, NE, USA) along with conductivity, turbid-ity, depth (pressure), temperature and pH (XRX 620, RBR, Ottawa, ON, Canada). Dissolved oxygen was simultaneously monitored online with normal and trace micro-optodes (types PSt1 and TOS7, Presens, Regensburg, Germany) with detection limits of 125 and 20 nM, respectively, and a response time of 7 s (Kirfet al., 2014).
Water samples for dissolved methane analysis were retrieved from distinct depths with a Niskin bottle. Serum bottles (120 ml) were filled completely without bubbles or headspace through a gas-tight outlet tubing allowing water to overflow. Solid copper chloride [Cu(I)Cl] was immediately added in excess to the water samples and the bottles were crimped. Before analysis, a 30 ml headspace was set with N2 and after overnight equilibration methane concentrations were measured in the headspace with a gas chromatograph (GC; Agilent 6890 N, Agilent Technologies, Santa Clara, CA, USA) equipped with a Carboxen 1010 column (30 m × 0.53 mm, Supelco, Bellefonte, PA, USA) and a flame ionization detector.
Methane concentrations in the water phase were back-calculated according to (Wiesenburg and Guinasso, 1979). Stable carbon isotopes of methane were determined in the same headspace by isotope ratio mass spectrometry with a trace gas instrument (T/GAS PRE CON, Micromass UK Ltd., Wilmslow, UK) coupled to a mass spectrometer (GV Instru-ments, Manchester, UK; Isoprime, Stockport, UK).
Isotopic ratios are given inδ-notation relative to the Vienna Pee Dee Belemnite reference standard.
Oxygen, PAR, methane concentration and methane isotope profiles for the sampling campaign in October 2014 are shown in Supplementary Figure 1. Geo-chemical profiles from other Lake Rotsee campaigns are reported in Oswaldet al.(2015).
Lake Rotsee methane oxidation rates
Methane oxidation rates were measured in incuba-tions set up in October 2014, with water from the 7 m depth (oxycline), and from 8 m depth (with no detectable oxygen). Water was collected with a Niskin bottle and filled into sterile 1 l Schott bottles without a headspace, closed with butyl stoppers and kept cold and dark until further handling. In the laboratory, 120 ml was distributed into 160 ml serum bottles in an anoxic (N2-containing) glove box (Iner Tec, Grenchen, Switzerland), closed with butyl stoppers and crimped. Each incubation was supple-mented with 13C-labeled methane (99 at%, Campro Scientific, Berlin, Germany) and 12C-methane to
2125
reach 2 bar overpressure, resulting in ~ 1.8 mmol l−1 CH4 in the water phase and 50 at% 13C labeling percentage. For comparison, in situ methane con-centrations at 7 and 8 m depth were ca. 15 and 35μmol l−1 (Supplementary Figure 1). Duplicate bottles were incubated at 6 °C under dark and light conditions along with a control (sterile filtered lake water). Methane oxidation was monitored during an incubation period of 7 days as production of 13CO2. Anoxically withdrawn water samples (2 ml) were transferred into 6 ml Exetainers (Labco, Lampeter, UK), fixed with 200μl zinc chloride (50% w/v) and acidified with concentrated H3PO4(100μl). Isotopic ratios of CO2were determined in the headspace with a preparation system (MultiFlow, Isoprime) coupled to an isotope ratio mass spectrometry (Micromass, Isoprime). Subsequently, methane oxidation rates were calculated as described previously (Oswald et al., 2015). These rates are shown in Supplementary Figure 1. As these incubations were unamended (apart from methane addition), aerobic methane oxidation in these incubations was presumably sus-tained solely by oxygenic photosynthesis (Milucka et al., 2015; Oswald et al., 2015). At selected time points, sub-samples were also taken for catalyzed reporter deposition fluorescence in situ hybridiza-tion (CARD-FISH) analysis. These data are shown in Supplementary Figure 3. Nanometer-scale secondary ion mass spectrometry (NanoSIMS) and metagenome analyses reported for Lake Rotsee (shown in Figure 1 and Supplementary Figure 2) were performed on samples collected on a previous sampling campaign in August 2013 (rates and other data from this campaign are reported in Oswaldet al. (2015)).
Lake Zug nitrate addition experiment
The sampling campaign was carried out in October 2013. Water samples from the anoxic 160 m depth were collected with a Niskin bottle, filled into sterile Schott bottles, closed with a stopper and stored as described above. The water was distributed into sterile 160 ml serum bottles (a 120 ml) in an N2 glove box (Mecaplex, Grenchen, Switzerland) as described in detail in Oswaldet al.(2016).13C-labeled methane (99 at%, Campro Scientific) was supplied at a ~ 20%
labeling percentage. A 2 bar methane overpressure was set using12C-methane. One set of duplicate bottles received no further addition and served as a control and one set of duplicate bottles was amended with
15NO3− (from a sterile anoxic 100 mmol l−1 stock solution) to a final concentration of 50μmol l−1. Bottles were incubated in the dark underin situtemperatures (~5 °C) for 16 days. At regular intervals, bottles were subsampled for 13CO2 measurements in order to determine methane oxidation rates. For this, anoxically withdrawn water samples (2 ml) were transferred into 6 ml Exetainers, fixed with zinc chloride and acidified with concentrated H3PO4. Isotopic ratios of CO2were determined in the headspace using a Finnigan Gas-Bench II attached to an isotope ratio mass spectrometer
(IRMS; Finnigan Delta Plus, Thermo Fisher Scientific, Waltham, MA, USA). Subsequently, methane oxida-tion rates were calculated as described previously (Oswald et al., 2015). At selected time points, sub-samples were also taken for CARD-FISH and nano-SIMS analyses. An early time point (T= 2d) was analyzed by nanoSIMS to obtain data for the calcula-tion of methane uptake rates reported in Table 1. FISH and nanoSIMS images from Lake Zug nitrate incuba-tion (Figure 1; Supplementary Figure 6) originate from the last time point of the incubation (T= 16 d). The sample for metagenome analysis (sample Z3) was also taken at this time point. Additionally, anin situwater sample from 160 m was also used for metagenome analysis (sample Z1). During this sampling campaign, no incubations with added oxygen were performed.
O2-supplemented incubations referred to in this manuscript were only performed during a sampling campaign in June 2014 and are described in detail in Oswald et al. (2016), where also the corresponding geochemical profiles and methane oxidation rates from relevant depths and incubations are reported.
Briefly, O2-supplemented incubations were set up as described above, with the difference that instead of nitrate, sterile air was injected to the incubations to reach final O2concentrations of ca. 80μmol l−1(‘low O2’) and ca. 200μmol l−1 (‘high O2’), respectively.
Incubations were subsampled at regular intervals for methane oxidation rates, CARD-FISH and nanoSIMS analyses. The CARD-FISH and nanoSIMS analyses shown in Figure 1 were performed on samples taken from 160 m incubation afterT= 2 d. The sample for metagenome analysis was taken at the last time point of the‘low O2’160 m incubation (T= 11 d).
Catalyzed reporter deposition fluorescencein situ hybridization
Formaldehyde- (2% (v/v) final concentration) fixed water samples were incubated for 30 min at room temperature before being filtered onto polycarbonate GTTP filters (0.2μm pore size; Merck Millipore, Darmstadt, Germany). For nanoSIMS analysis, samples were filtered onto Au or Au/Pd-coated GTTP filters (0.2μm pore size). Permeabilization with lysozyme, peroxidase inactivation, hybridization with specific oligonucleotide probes labeled with horseradish per-oxidase in combination with tyramide signal amplifi-cation (Oregon Green 488) and DAPI counter staining was performed as described previously (Pernthaler et al., 2002). An overview of probes used (Biomers, Ulm, Germany) is included in Supplementary Table 2.
For cell counts and biovolume determinations, one filter was analyzed for each sample. Hybridized filaments (using probe Mgamma669) were enumerated in randomly selected fields of view with a confocal laser scanning microscope (SP5 DMI 6000, Leica, Wetzlar, Germany). For biovolume calculations, length and width of415 filaments in410 fields of view were then measured directly in confocal micrographs using LAS AF Lite software (Leica). Values for the cell counts
2126
and methane uptake rates of unicellular gamma-MOB cells were taken from Oswaldet al.(2015) and Oswald et al.(2016).
Nanometer-scale secondary ion mass spectrometry Areas of interest containing positive CARD-FISH hybridization signals were marked with a laser micro-dissection microscope (DM 6000, Leica Micro-systems, Mannheim, Germany). Laser-marked areas were analyzed by nanoSIMS (NanoSIMS 50 l, Cameca, Paris, France) at the MPI Bremen as described previously. For Lake Rotsee (light incubation, 9 m depth), 12 and 26 filaments were analyzed in five fields of view after 2 and 7 days of incubation,
respectively. For the Lake Zug low and high O2
addition experiments, 19 and 13 filaments were measured in 9 and 7 fields of view, respectively, after 2 days of incubation. For the Lake Zug nitrate addition incubation, 6 filaments were measured in 5 fields of view after 2 days of incubation and 7 filaments were measured in 5 fields of view after 16 days of incubation. Obtained secondary ion images were drift corrected, accumulated and processed with Look@Na-noSIMS (Polereckyet al., 2012).
Biovolume and carbon assimilation rates
The biovolume of individual Crenothrix filaments was calculated from their measured length and width
Figure 1 Methane-dependent growth ofCrenothrix in Lake Rotsee and Lake Zug. (a)Crenothrixin the Lake Rotsee oxic incubation visualized by CARD-FISH (green; counterstained by DAPI in blue) with a specific probe Creno445 (Stoeckeret al., 2006). A small coccoid cell targeted by the probe (marked by the asterisk) might represent a gonidial cell, whichCrenothrixis reportedly capable of producing (Völkeret al., 1977). (b) The corresponding13C/12C nanoSIMS image shows homogeneous13C enrichment throughout the cell filament.
The small coccoid cell is also significantly enriched, albeit less. (c) The corresponding32S/12C nanoSIMS image showing distribution of organic material on the filter. (d) PutativeCrenothrixfilaments in the Lake Zug oxic incubation visualized by DAPI (blue) and CARD-FISH (green) with probe Mgamma669. (e) Corresponding13C/12C and (f)32S/12C nanoSIMS images. Note the fragmented nature of theCrenothrix filaments and the attached small (unidentified) bacteria. (g) PutativeCrenothrixfilaments in the Lake Zug anoxic incubation visualized by DAPI (blue) and CARD-FISH (green) with probe Mgamma669. (h) Corresponding13C/12C and (i)32S/12C nanoSIMS images.
2127
Table1OverviewofmethanecarbonuptakeratesbyCrenothrixandunicellulargamma-MOBinLakeRotseeandLakeZug 13Cat%a (n)Avgbiovolume (CARD-FISH-based)b (μm3;n) Methane-Cuptakepercellc (fmolcellavg−1d−1)Cellcountd (cellml−1)Totalpopulation biovolumee (μm3ml−1)
Methane-Cuptake perpopulationf (μmoll−1d−1) LakeRotsee(oxic) Crenothrix(Mgamma669)NA85±8.3(59)147.7±26.3g1.2E+041.0E+061.73g Crenothrix(Creno445)22.00±4.8(17)73.7±8.4(51)128.0±22.89.2E+036.8E+051.18 Othergamma-MOBh28.77±4.14.210.6±0.92.6E+041.1E+050.27 LakeZug(oxic) Crenothrix(Mgamma669;lowO2)9.26±1.7(19)32.5±5.5(20)38.1±6.91.1E+033.5E+040.041 Crenothrix(Mgamma669;highO2)8.68±1.9(10)i32.5±5.5(20)35.3±7.81.1E+033.5E+040.038 Othergamma-MOB(lowO2)j10.39±3.14.25.7±1.26.8E+042.9E+050.39 Othergamma-MOB(highO2)j12.13±3.754.26.9±1.66.8E+042.9E+050.47 LakeZug(anoxic) Crenothrix(Mgamma669)13.27±4.9(6)49.7±20.3(15)74.2±26.6(6)0.4E+032.0E+040.03 Othergamma-MOBNANANANANANA Abbreviations:CARD-FISH,catalyzedreporterdepositionfluorescenceinsituhybridization;MOB, methane-oxidizingbacteria;n,numberofanalyzedcells,NA,notanalyzed. aCalculatedasanaverage(±s.d.)ofthe13C/12Cratiosofindividualregionsofinterest(i.e.,cells)determinedbynanoSIMS. bCalculatedfromCARD-FISHdataasanaveragebiovolume(±s.d.)usingMgamma669orCreno445probe(Crenothrix)andMgamma84+705probes(othergamma-MOB). cCalculatedasfollows:datafromcolumnawereconvertedinto13Cexcessinfmolpercell(ofagivenaveragebiovolume;cellavg)usingtheavgcellbiovolumereportedincolumnbandaconversion factorof6.4fmolCμm−3(Musatetal.,2008).Thenumberswerecorrectedforlabelingpercentageandincubationtime. dCountedfromthesamefiltersfromwhichavgbiovolumes(columnb)wereobtained.Astheboundariesbetweenindividualcellswithinthefilamentwereoftennotrecognizable,onlyhybridized filamentswerecounted.Cellcountsrefertocellabundancesatthestartofeachincubationandthusdonotaccountforincreaseofcellabundancesduringtheincubationperiod. eCalculatedasfollows:datafromcolumnbwereupscaledusingdataincolumnd. fCalculatedasfollows:datafromcolumncwereupscaledusingdatafromcolumnd. gAssumingthesame13CenrichmentasdeterminedwiththeprobeCreno445onthesamesample. hAccordingtoOswaldetal.,2015. iInthissample,threeanalyzedfilamentshad13C/12Co0.015andwerenotincludedintheanalysis. jAccordingtoOswaldetal.,2016. CalculationsarebasedonincubationsfromLakeRotsee(oxic,2013)andLakeZug(oxicandanoxic,2013,2014;seeSupplementaryTable4forsampledetails).
2128
by assuming a cylindrical shape. The length and the width of filaments were determined from the CARD-FISH images that were used for cell counting. Due to the varying length of filaments, an average biovolume of Crenothrix was calculated and is reported in Table 1. The ‘average biovolume determined from CARD FISH’ was calculated as an average of biovolumes of individual filaments hybridized with a Crenothrix-targeting probe (Mgamma669 or Creno445) at the start of the respective incubation and is reported with the s.d. ‘Total’ Crenothrix biovolume reported in Table 1 and Supplementary Figure 4 was obtained by multiplying the average filament biovolume by the number of filaments per ml of water. For comparison, the biovolume of unicel-lular gamma-MOB cells was calculated from total cell counts and by assuming an average spherical cellular diameter of 2μm.
Cellular 13C at % were calculated from 13C/12C values of individual ROIs (regions of interest).
Regions of interests were drawn to outline single Crenothrix cells (for example, Figures 1h and i), whole filaments (Figures 1e and f) or parts of filaments (Figures 1b and c). Both the background (cell-free polycarbonate filter in the same field of view) and the13C enrichment of all cells in every field of view was evaluated and compared for all measure-ments. Rates of methane carbon uptake (fmol C cellavg−1d−1) of Crenothrix and unicellular gamma-MOB were calculated from the 13C excess of the measured cells using a conversion factor of 6.4 fmol Cμm−3reported in Musatet al.(2008). These uptake rates were corrected for the labeling percentage and the incubation time. The methane uptake rates were calculated only for filamentous cells, which were stained with the Creno445 or Mgamma669 probe.
Hybridized single cells (such as in Figures 1a–c) were not considered in the calculation.
DNA extraction, 16S rRNA gene amplicon sequencing and analysis
Twoin situwater samples from Lake Rotsee were used for 16S rRNA gene amplicon sequencing. One was collected from the oxycline (9 m depth) during a campaign in August 2013 and the other from anoxic water (8 m depth) during a campaign in October 2014 (Supplementary Table 4). Volumes of ca. 250 ml were filtered onto polycarbonate Nuclepore Track-Etched Membrane filters (0.2μm pore size; Whatman, Maid-stone, UK). Filters were stored at −80 °C until DNA was extracted with the UltraClean Soil DNA Isolation Kit (MoBio Laboratories, Carlsbad, CA, USA). Extrac-tion procedure was performed according to manufac-turer’s instructions with the following adjustment:
vortexing with the Bead Solution was reduced to 30 s with subsequent incubation on ice (30 s), and this cycle was repeated four times.
The V3–V4 regions of the 16S rRNA gene were targeted with primer pair 341 F (5′-CCTACGGGNG GCWGCAG-3′) and 805 R (5′-GACTACCAGGGTATC
TAATC-3′). The forward primers contained unique identifier sequences at the 5’-end for each sample to allow for multiplex sequencing. Ten separate PCR reactions (25μl volume) were set up for each sample including both forward and reverse primers (500 nM each), deoxyribose nucleotide triphosphates (dNTPs;
800μM), 1 × Taq reaction buffer, Taq DNA polymer-ase (0.25 U) and DNA extracts of the respective samples (0.5–1μl). The reactions proceeded as follows: initial denaturation (3 min at 95 °C), 25 cycles of denaturation (30 s at 95 °C), annealing (30 s at 54 °C) and elongation (90 s at 72 °C); and final elongation (10 min at 72 °C). Parallel reactions were combined and purified with the QIA quick PCR Purification Kit (Qiagen, Hilden, Germany) following manufacturer’s instructions, with a final elution in 1 × TE buffer (30μl; 10 mM Tris-HCl (pH 8.0)+1 mM
EDTA). The DNA was further purified with a gel using SYBR Green I Nucleic Acid Gel Stain (Invitro-gen, Carlsbad, CA, USA) followed by gel extraction with QIAquick Gel Extraction Kit (Qiagen) according to the manufacturer’s protocol. Extract concentra-tions were measured fluorometrically using the Qubit dsDNA HS Assay Kit and the Qubit 2.0 Fluorometer (Invitrogen). Illumina sequencing was performed on the amplicons at the Max Planck-Genome Centre (Cologne, Germany).
16S rRNA gene amplicon paired-end reads were trimmed (right end only, trim quality threshold = 10) and merged (20 bases minimum overlap) using BBmap software version 35.43 (sourceforge.net/projects/
bbmap). Reads were then separated by barcode and trimmed (minimum length = 300, maximum homo-polymer length = 8, maximum number of ambiguous bases = 0, minimum average quality score allowed over 50 bp window = 20) using mothur v.1.36.1 (Schloss et al., 2009). The separated reads were processed using SILVAngs and standard parameters (Quastet al., 2013).
Lake metagenome sequencing and assembly
Two in situ water samples (Lake Rotsee, 9 m depth, August 2013 (sample R1) and Lake Zug, 160 m depth, October 2013 (sample Z1)) and four end time points of incubations (Lake Rotsee, O2-supplemented (sample R2), Lake Rotsee, light (sample R3), Lake Zug, low O2 -supplemented (sample Z2), Lake Zug, anoxic, nitrate-supplemented (sample Z3); see Supplementary Tables 3 and 4 for additional sample information) were analyzed by Illumina sequencing. The following water volumes were filtered onto polycarbonate Nucleopore Tracked-Etched membrane filters (0.2μm pore size; Whatman) and stored at−80 °C: 250 ml forin situsamples (R1 and Z1), 50 ml for Lake Rotsee incubations (R2 and R3) and 40 ml for Lake Zug incubations (Z2 and Z3). DNA was extracted from cut-up filters using the PowerSoil DNA isolation kit according to manufacturer’s instructions (MoBio Laboratories). DNA from lake Zug was frag-mented by sonication (MiSeq: 600–700 bp; HiSeq2500:
300 bp) using a Covaris S2 sonicator (Covaris, Woburn, MA, USA). The library was prepared using Ovation
2129
Ultra Low Library Systems V1 (for MiSeq) or V2 (for HiSeq2500) kits (NuGEN Technologies, San Carlos, CA, USA) and paired-end sequencing (2 × 300 or 2 × 150 bp) was performed using the Illumina MiSeq (2 × 300 bp) or HiSeq2500 (2 × 150 bp) platform (Illumina Inc., San Diego, CA, USA). DNA from Lake Rotsee was fragmented by sonication (350 bp) using a Covaris S2 sonicator (Covaris), the library was prepared using NEBNext Ultra DNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA, USA) and paired-end sequencing (2 × 150 or 2 × 100 bp) was performed using the Illumina HiSeq2500 or 3000 platform (Illumina Inc.). Both MiSeq and HiSeq sequencing was performed by the Max Planck-Genome-centre, Cologne, Germany (http://mpgc.mpipz.mpg.de/home/;
Supplementary Table 3).
Sequences were quality checked using FastQC (Andrews, 2010) and trimming, as well as adapter removal was done using Trimmomatic 0.32 and parameters MINLEN:20 ILLUMINACLIP:TruSeq3-PE.fa:2:30:10 LEADING:3 TRAILING:3 SLIDING-WINDOW:4:15 MINLEN:50 (Bolger et al., 2014).
Metagenome assembly of sequences from the Lake Zug incubation (anoxic, nitrate-supplemented (Z3;
Supplementary Tables 3, 4)) was performed using SPAdes 3.5.0 (Bankevichet al., 2012) with mismatch corrector enabled and default parameters.
Sand filterCrenothrixmetagenome sequencing and assembly
Samples containing high proportions ofC. polyspora filaments were taken from the backwash water of rapid sand filters of the Wolfenbüttel waterworks (Germany), which treats a mixture of oxic and anoxic groundwater. During sampling,Crenothrixfilaments were retained from 600 to 850 liters of backwash water by either sedimentation or filtration through a fine-mesh sieve (200 or 400μm). One sample was collected in 2004 (on 21 June; sample C) and was incubated with 500μmol l−1 ammonium for 212 h.
The second sample was collected in 2005 (10 October, sample B) and was incubated at different methane concentrations for 24 h. It should also be noted that earlier we deposited one additional partial and unpublished Crenothrix genome from a sand filter sample from the Wolfenbüttel waterworks at IMG (genome ID 3300005627). We did not analyze that older genome sequence in the course of the present study, because it originated from the same site but had been sequenced less deeply than the two sand filterCrenothrixgenomes described here.
After the incubations, samples B and C were frozen at −20 °C and DNA was extracted in 2016 using a phenol chloroform protocol (Zhou et al., 1996) including two bead-beating steps. Paired-end sample libraries were prepared using Illumina Nextera DNA Library Preparation Kit (Illumina Inc.) and sequenced at Aalborg University (Denmark) using an Illumina MiSeq with MiSeq Reagent Kit v3 (2 × 301 bp; Supplementary Table 3). Paired-end
reads were imported to CLC Genomics Workbench v. 8.0 (CLCBio, Aarhus, Denmark) and trimmed using a minimum phred score of 20, a mini-mum length of 50 bp, allowing no ambiguous nucleotides and trimming off Illumina sequencing adaptors if found. All trimmed paired-end metagen-ome reads were assembled using CLC’s de novo assembly algorithm, using a kmer of 63 and a minimum scaffold length of 1 kbp.
Metagenome binning, reassembly and annotation Binning of contigs of the Lake Zug metagenomic assembly (sample Z3, Supplementary Table 3) was performed by exploiting differential contig coverage from three sequenced metagenomic data sets: Z1 (Lake Zug, in situ), Z2 (Lake Zug, O2-supplemented incuba-tion) and Z3 (Lake Zug, anoxic, nitrate-supplemented incubation) as described previously (Albertsen et al., 2013) and implemented in the mmgenome R package (http://madsalbertsen.github.io/mmgenome/; Karstet al., 2016). Only contigs longer than 500 bp were used and the average coverage of each contig was computed directly using BBmap 35.43 (http://sourceforge.net/
projects/bbmap/) with default parameters. Prodigal 2.60 (Hyattet al., 2010) in metagenomic mode (-p meta) and standard parameters was used to predict open reading frames, which were translated to amino-acid sequences and subsequently searched for using HMMER 3.1b (Eddy et al., 2013) against a set of 107 hidden markov models of essential single-copy genes (Dupont et al., 2012) using default settings and trusted cutoff (-cut_tc) enabled. Protein sequences coding for essential single copy genes were searched against NCBI non-redundant database (retrieved in August 2015) using BLASTP (Camachoet al., 2009) and an e-value cutoff of 10−6. The taxonomy (class level) of each essential single-copy gene was assigned using MEGAN5 (Husonet al., 2011; with the previously generated BLASTP xml file as input) and the mmgenome script‘hmm.majority.vote.pl’.
Bowtie2 (Langmead and Salzberg, 2012) with standard settings was used to map reads to contigs and the number of paired-end connections between separate contigs was calculated from the SAM file using the mmgenome script‘network.pl’.
Differential coverage of contigs between the two sand filterCrenothrixmetagenomes (Supplementary Figure 8) and between the Lake Zug metagenomes (Supplementary Figure 7), as well as paired-end connections between separate contigs were used to extract genomic bins from the metagenome using the mmgenome R package (http://madsalbertsen.github.
io/mmgenome/; Karst et al., 2016). Reads used for the initial assembly were mapped to the binned contigs using BBmap of the BBmap package 35.43 (http://sourceforge.net/projects/bbmap/) using strin-gent settings (approximate minimum identity = 0.98) or CLC (sand filterCrenothrix). Mapped reads were reassembled (only for the lacustrine Crenothrix) using SPAdes 3.5.0 (Bankevich et al., 2012) with mismatch corrector enabled and default parameters.
2130
Quality of the reassembled bins was assessed using CheckM 1.05 running the lineage-specific workflow (Parkset al., 2015). Annotation of theCrenothrixD3 draft genome was performed using RAST (Azizet al., 2008). CDS prediction and automated pre-annotation of the two Wolfenbüttel sand filter Crenothrix genome sequence bins were performed using the PROKKA pipeline (Seemann, 2014) with an in-house extended protein reference database. The annotation of key metabolic pathways was manually refined.
The Whole Genome Shotgun project of lacustrine Crenothrix sp. D3 has been deposited at DDBJ/ENA/
GenBank under the accession MBQZ00000000.
The version described in this paper is version MBQZ01000000. Reads (Lake Zug and Lake Rotsee) have been deposited at the Sequence Read Archive under BioProject PRJNA325574. The two sand filter Crenothrixmetagenomic assemblies are available in the European Nucleotide Archive (ENA) under the study accession number PRJEB19189.
Phylogenetic analyses
Full-length amino-acid sequences of bacterial PmoA and AmoA protein sequences were retrieved from the Integrated Microbial Genomes database (IMG-ER;
Markowitzet al., 2009) using Pfam family PF02461.
Previously published protein sequences of‘unusual’
PmoA of C. polyspora (accession ABC59822–
ABC59827; Stoecker et al., 2006), partial PmoA of C. fusca(accession ABL64049; Vigliottaet al., 2007) AmoA sequences belonging to CandidatusNitrospira nitrosa, (accession CUS31358; van Kessel et al., 2015) as well as Candidatus Nitrospira inopinata (accession CUQ66826; Daims et al., 2015) were added to the reference set. After removing duplicate sequences, protein sequences were aligned using Clustal Omega 1.2.0 (Sieverset al., 2011) and default parameters. A phylogenetic tree (135 taxa) was calculated using RAxML 8.2.6 (Stamatakis, 2014) and parameters: -f a -k -x 48020621 -p 6809427 -N 100 -T 8 -m PROTGAMMAWAG.
Partial Crenothrix 16S rRNA gene sequences were retrieved from the Crenothrix draft genomes using RNAmmer 1.2 (Lagesenet al., 2007), aligned using the SILVA incremental aligner (SINA) 1.2.11 (Pruesseet al., 2012) and imported to the SILVA SSU NR99_123 database (Quast et al., 2013) using ARB 6.1 (Ludwig et al., 2004). Phylogenetic trees of the 16S rRNA gene sequences were calculated using RAxML 7.7.2 inte-grated in ARB with the GAMMA model of rate heterogeneity and the GTR substitution model with 100 bootstraps.
Results and discussion
Crenothrixin Lake Rotsee and Lake Zug
To investigate the potential occurrence of filamen-tousCrenothrix bacteria in two stratified lakes and their involvement in the lacustrine methane cycle,
we first recorded geochemical evidence for methane oxidationin situ. Concentration profiles recorded in Lake Rotsee and Lake Zug over the course of 3 years suggested a zone of methane consumption that persistently coincided with the oxycline (profiles from Lake Rotsee 2013 are shown in Oswald et al.
(2015), from 2014 in Supplementary Figure 1;
profiles from Lake Zug 2012, 2013 and 2014 are shown in Oswald et al., 2016). Concurrently, incubations with 13CH4 confirmed high rates of methane oxidation at the oxycline (Oswald et al., 2015, 2016; Supplementary Figure 1). These incuba-tions were set up under both oxic and anoxic conditions. In Lake Rotsee, oxic incubation condi-tions were obtained either by addition of air or solely by incubation of anoxic water in the light. In the latter case, aerobic methane oxidation was presum-ably sustained by oxygenic photosynthesis (Milucka et al., 2015; Oswaldet al., 2015). In Lake Zug, oxic incubations were solely supplemented with air and incubated in the dark. These different incubation set ups reflected the different nature of the two lakes, Lake Rotsee has a shallow, sun-lit oxycline, whereas the oxycline of Lake Zug is very deep and dark.
Additionally, anoxic Lake Zug incubations supple-mented with nitrate were also set up as Lake Zug had the appropriate environment to test for methane-dependent denitrification (Supplementary Table 4).
We then analyzed the microbial community at the Lake Rotsee oxycline by 16S rRNA gene amplicon sequencing in 2 consecutive years (2013 and 2014;
Supplementary Figure 2). Along with gamma-proteobacterial Methylococcaceae (Methylobacter, Methylocaldum, Methylomonas and Methyloglobu-lusspecies), CABC2E06 (an uncultured Methylococ-cales clone; Wanget al., 2012; Quaiseret al., 2014), and the marine methylotrophic group, also sequences belonging to Crenothrix were retrieved.
On the basis of the number of recovered sequences, Crenothrix-related organisms were 2–5-fold less abundant than Methylococcaceae and comprised 0.06–0.1% of the total bacterial sequences in situ.
However, it is possible that the true abundance of Crenothrix in situ was higher than what the 16S rRNA gene abundances suggest, as, for example, DNA extraction biases might strongly select against these thickly sheathed microorganisms.
We could additionally confirm the presence of Crenothrix in both lakes by CARD-FISH with two oligonucleotide probes reported to targetCrenothrix, Mgamma669 and Creno445 (Eller et al., 2001;
Stoeckeret al., 2006). The more specific oligonucleo-tide probe Creno445 bound only sporadically, when the hybridization stringency was strongly reduced (Supplementary Figure 3). On the other hand, the Mgamma669 probe hybridized most of the conspic-uous filaments in all analyzed samples from both lakes (in situwater as well as incubations, Figure 1;
Supplementary Figure 3) even though some fila-ments did not hybridize even with this more general
2131
probe (for example, Supplementary Figures 3a, b).
With both probes, we observed two hybridized cell morphotypes—filaments and single round cells (Figure 1; Supplementary Figure 3). Both morpho-types have been observed for Crenothrix spp.
previously and it has been proposed that the smaller round cells represent reproductive cells that bud from the ends of vegetative cell filaments (Cohn, 1870; Völker et al., 1977). However, given the compromised specificity of the Creno445 probe at low stringency and the broad specificity of the Mgamma669 probe, it is also possible that the hybridized single cells represented other gamma-MOB, reportedly targeted by the Mgamma669 probe (for example,MethylobacterorMethylomonas; Eller et al., 2001). Therefore, the here-reportedCrenothrix cell counts and biovolumes are solely based on counts of Creno445- or Mgamma669-hybridized filaments and thus represent conservative estimates.
Overall, in all analyzed incubations from both lakes total Crenothix biovolumes increased over time (Supplementary Figure 4b). This confirms that Crenothrixwas growing under both oxic and anoxic conditions.
Whereas unicellular gamma-MOB had consis-tently cell sizes of ca. 2μm, the individual cells in Crenothrix-like filaments reached an average length of ca. 5μm (Figure 1; Supplementary Figures 3a and 5a). The average length and width of Lake Rotsee Crenothrix filaments was ca. 45 and ca. 1.5μm, respectively, with individual filaments reaching 4100μm length (Supplementary Figure 3). Fila-ments were often intertwined and bunched together, as observed previously (Cohn, 1870; Völker et al., 1977). In Lake Rotsee, the biovolume of Crenothrix was about eight-fold higher than that of unicellular gamma-MOB at depths corresponding to the highest observed methane oxidation rates (in 2012 and 2013;
Supplementary Figure 4a). Only in 2014 unicellular gamma-MOB biomass contribution was higher than that of Crenothrix (Supplementary Figure 4).
We speculate that these differences might be con-nected to the complex life cycle of Crenothrix (Supplementary Discussion). In Lake Zug, the fila-ments were shorter but more consistent in terms of length, reaching an average length and width of ca.
28 and 1.4μm (in 2013) and ca. 20 and ca. 1.4μm (in 2014), respectively.
Methanotrophic growth ofCrenothrix
To confirm that the observed cell growth (that is, increase in cell numbers and biovolume over time;
Supplementary Figure 4b) was methane-derived, samples from the 13CH4-supplemented incubations were further analyzed by nanoSIMS. Filamentous bacteria hybridized with the Mgamma669 probe consistently constituted the highest 13C-enriched population in all three investigated incubations (Lake Rotsee oxic, Lake Zug oxic and Lake Zug anoxic; Figure 1; Supplementary Figure 5). The 13C
enrichment confirmed that 13CH4 was assimilated into cell biomass, such as is common for gamma-proteobacterial methanotrophs (Trotsenko and Murrell, 2008). In some of the images, fragmentation of filaments into single vegetative cells was apparent, even though the uptake of 13C appeared homoge-nously spread throughout the whole filament. In both lakes, Crenothrix filaments appeared to be colonized by other non-identified bacteria, which did not show comparably strong enrichment in13C and might thus represent heterotrophic epibionts (Figure 1). In contrast, the single round cells (hybridized with Mgamma669 probe) were similarly enriched in 13C as the Crenothrix filaments (Figures 1a–c), supporting the speculation that these cells belong to methanotrophic bacteria and might potentially represent reproductiveCrenothrixcells.
In the Lake Rotsee oxic incubation, the uptake of methane-derived carbon byCrenothrixfilaments was comparable to that of‘classical’unicellular gamma-MOB (13C enrichment of 22 ± 4.8 at % and 29 ± 4.1 at
%, respectively; Table 1; Figure 1; Supplementary Figure 2). However, due to its larger biovolume Crenothrix assimilated ca. 4–6-fold more methane than the‘classical’gamma-MOB in the same incuba-tion (1.73 or 1.18μmol methane l−1d−1 and 0.27μmol methane l−1d−1, respectively; Table 1).
These numbers are based on average filament biovolumes and cell counts determined by CARD-FISH at the beginning of the incubation and do not take into account any increase in cell numbers over time, as the incubation conditions might have differently affected the growth of the different MOB. However, even if we take into account the increase of cell numbers over time, overall contribu-tion ofCrenothrixto methane uptake in Lake Rotsee was still higher than that of the unicellular gamma-MOB, even though the difference was not so pronounced (ca. 1.4 higher based on Tend cell counts).
Crenothrixfilaments in Lake Zug oxic incubations were also active and assimilated methane at rates of ca. 0.04μmol methane l−1d−1 (Table 1;
Figures 1d–f). This is much lower than the overall contribution ofCrenothrix in Lake Rotsee, which is largely due to their lower abundance (1.1E+03 cells per ml) and smaller average biovolume (ca.
30μm3).
Additionally, Crenothrix was also active in our anoxic denitrifying incubations where not enough oxygen was present to account for measured methane oxidation rates (2.7μmol l−1d−1 13CO2
produced in 15NO3-supplemented incubation, a ca.
10-fold increase compared to control incubation without any added electron acceptor (0.234μmol l−1d−1 13CO2 produced)). The methane-dependent growth under oxygen-deficient conditions was evi-denced as cell biomass enrichment in both13C (from
13C-CH4; Figures 1g–i) and 15N (from 15N-nitrate;
Supplementary Figure 6), even though the methane uptake rates were somewhat lower (0.03μmol
2132